Effects of dimensions on the sensitivity of a conducting polymer microwire sensor Cheng Luo à , Anirban Chakraborty a Institute for Micromanufacturing, Louisiana Tech University, 911 Hergot Avenue, Ruston, LA 71272, USA article info Article history: Received 26 August 2008 Received in revised form 20 November 2008 Accepted 24 November 2008 Keywords: Conducting polymers Microwire sensors Surface-to-volume ratio Sensitivity Intermediate-layer lithography abstract It is commonly considered that the sensitivity of a microsensor increases with its increasing surface -to- volume ratio. However, it is not exactly clear how the surface-to-volume ratio affects the sensitivity of a conducting polymer microsensor. The change in any of the three geometrical dimensions (i.e., length, width and thickness) of a microsensor changes the surface-to-volume ratio. In designing a microsensor of desired sensitivity, it is important to know the effect of each dimension on the sensitivity for properly defining the sizes and shapes of the microsensor. As such, in this work, we have investigated the effects of each individual dimension on the sensitivity of a conducting polymer microwire sensor. Polypyrrole (PPy) and Poly (3,4-dimethlydioxythiophene) poly(styrenesulfonate) (PEDOT–PSS) microwire sensors of different dimensions were fabricated using an intermediate-layer lithography (ILL) method. They were further employed to detect methanol and acetone vapors at concentrations in the range of 0.6–7.1 parts per thousand (ppt). The corresponding three relationships between the three geometrical dimensions and the sensitivities were found using a statistical program, SAS. From the point view of surface-to- volume ratio, the thickness should affect the sensitivity much more than the other two dimensions. However, the th ree relationships indicate that the effects of the three geometrical dimensions on the sensitivity of a microwire sensor vary with the conducting polymer materials and the targets to detect. In other words, which dimension has more effects on sen sitivity is case-dependent. Results presented in this work can be potentially used to aid in the design of conducting polymer microwire sensors of high sensitivity. & 2008 Elsevier Ltd. All rights reserved. 1. Introduction Conducting polymers have received much attention since their discovery in 1977. Applications of conducting polymer micro- systems span from electronic devices to biological and chemical sensors. Conducting polymers offer some unique advantages like low weight, easy tailoring of properties and a wide spectrum of applications [1–3]. The conducting polymers produce changes in color, mass, work function and conductivity when exposed to different chemicals [4]. A commonly used sensing mechanism is through conductivity measurements of an exposed polymer film [5]. The corresponding operating principle is the resistance change of a conducting polymer film upon exposure to a particular chemical analyte. Most conducting polymers respond to the exposure of an analyte with a unique change in conductivity. This response is reversible with original behavior recovered as soon as the exposure is stopped. The adsorption/ desorption-related conductivity changes normally occur at room temperature. These so-called ‘‘chemiresistors’’ are easier to implement experimentally [6–9]. Two of the most commonly used conducting polymers are Polypyrrole (PPy) [8,10–18] and Poly (3,4-dimethlydioxythiophene) poly(styrenesulfonate) (PEDOT–PSS) [5,7,9,19]. These polymers have been used to sense for various chemical analytes including water vapor [5,9,10,18], volatile organic gases [5,7,8,11–13,16,18,19] (such as methanol, acetone, alcohol, and ethanol), industrial gases [14] (such as ammonia, NO x ,CO x ,SO 2 ,H 2 S, O 2 and H 2 ), glucose [15], and antigens [17]. Compared to film sensors, microsensors generally have exhibited higher sensitivity in detecting analytes of low concen- trations. It is normally considered that the higher sensitivity is induced by the higher surface-to-volume ratio of the micropat- terns. However, it is not exactly clear how the surface-to-volume ratio affects the sensitivity of a conducting polymer microsensor. A recently developed intermediate-layer lithography (ILL) enables us to properly fabricate conducting polymer microsensors. There- fore, in this work, PPy and PEDOT–PSS microwires of different dimensions have been fabricated using the ILL method and subsequently applied to detect methanol and acetone vapors of concentrations in the range of 0.6–7.1 parts per thousand (ppt). ARTICLE IN PRESS Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/mejo Microelectronics Journal 0026-2692/$ -see front matter & 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.mejo.2008.11.064 à Corresponding author. Current address: Department of Mechanical and Aerospace Engineering, University of Texas at Arlington, 500 W First Street, Arlington, TX 76019, USA. Tel.: +1817 272 7366; fax: +1817 272 5010. E-mail address: chengluo@uta.edu (C. Luo). Microelectronics Journal ] (]]]]) ]]]–]]] Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 The microwire response was also compared with the response of a square film (1 cm  1 cm). The corresponding relationships between the three geometrical dimensions and the sensitivities were found using a statistical program, SAS, to find the effects of each individual dimension. The outline of this work is as follows. Section 2 discusses the detection principle of the conducting polymer film and microwire sensors. In Section 3, the fabrication of the conducting polymer microwires of various dimensions is detailed with experimental results. The experimental setup for detecting methanol and acetone vapors is presented in Section 4. Section 5 compares the sensing results of conducting polymer film and microwire sensors. In Section 6, the effects of each individual dimension of a sensor on the sensitivity are addressed. This work is finally summarized and concluded in Section 7. 2. Sensing principles In this section, we first show that the sensitivity of a conducting polymer microsensor actually depends on the sensitivity of a unit block, and then discuss ways to increase the sensitivity of the unit block. As what has been done by many researchers [8,10–13,17,18], the sensitivity index (SI) is defined as (R exposure ÀR Base )/R Base , where R Base and R exposure represent the resistances of a sensor before and after exposure to a target, respectively. The SI indicates how large the sensor response is to a particular concentration, and is used as a measure of the sensitivity of the corresponding sensor. Geometrically, the sensing area of a film sensor can be modeled to be made up of multiple microwires of unit width, connected in parallel between the opposite edges at the electro- des. These microwires may be further divided into blocks of unit top area along the entire length (Fig. 1). These blocks of unit area may be regarded as individual ‘‘chemiresistor’’ elements with the base resistance r Base i;j , which respond to the various concentrations of analytes with unique changes in conductivity. These ‘‘chemir- esistor’’ blocks may be treated to be electrically connected in a serial fashion between the opposite electrodes. Let the resistance of the block be r 0 i,j upon exposure to an analyte. Therefore, the SI for this block would be (r 0 i,j Àr Base i;j )/r Base i;j . If the microwire is divided into ‘‘n’’ identical blocks, the total base resistance of a single microwire would be n  r Base i;j . Upon exposure to analyte, the resistance would be n  r 0 i,j . As there are ‘‘m’’ identical microwires connected in parallel, the overall base resistance would be (n/m)  r Base i;j and the resistance upon exposure to analyte would be (n/m)  r 0 i,j . The SI for the film sensor would be ð D R=RÞ Total ¼ f½ðn=mÞÂr 0 i;j Àðn=mÞÂr Base i;j =½ðn=mÞÂr Base i;j g ¼ð D R=RÞ Block . (1) It is observed from Eq. (1) that the SI of a film sensor equals that of a single unit block, which does not depend on how many unit blocks this film sensor has. In deriving Eq. (1) for a film sensor, it is assumed that the unit blocks have the same sensitivity. This assumption holds when the top surface of the film is much larger than the side surfaces. During the detection, a film sensor has five surfaces exposed to a target: the top and four side surfaces. The bottom surface interfaces with the substrate, and is not exposed to a target. Since the top surface is much larger than the four exposed side surfaces, most of the unit blocks only get exposed to a target through their top surfaces of unit area. That is, most blocks get the same exposure to a target. Furthermore, these unit blocks have the same geometry. Therefore, Eq. (1) is reasonably true. In the case of, for example, microwire sensors, the sizes of the top surface may be comparable with those of the side surfaces. Unit blocks located at the edges of the sensors get more exposure to a target than those in the central area of the sensor. The unit bocks may have different sensitivities. Accordingly, the assumption in deriving Eq. (1) may not hold. In this case, the surface-to-volume ratio should be considered to address the average sensitivities of the unit blocks. This ratio means that how much surface of a block, which has a unit volume, is exposed to a target. In principle, more exposed surface implies that the block should be more affected, having higher sensitivity. Consider a rectangular pattern, which has a length a, a width b, and a thickness t (Fig. 2). The sensing surface area is (a  b+2  a  t+2  b  t). The volume of the film is (a  b  t). Therefore, the surface-to-volume ratio is (1/t+2/a+2/b). It can be seen that this ratio increases with decrease in length, width and thickness. For a microsensor fabricated out of thin films, which normally have thicknesses ranging from tens of nanometers to several microns, the width and length are generally above 10 m m and much larger than the thickness. As such, 1/t is much larger than 2/a and 2/b. In other words, the changes of a and b do not affect the ratio much for a fixed t. For example, when t ¼ 1 m m, a ¼ 10 m m and b ¼ 10 m m, the ratio is 1.4 m m À1 . The reduction of both a and b by half yields a new ratio of 1.8 m m À1 , while the reduction of t by half leads to a new ratio of 2.4 m m À1 . In short, the thickness is the most important dimension among the three in affecting surface-to-volume ratio. In this work, we explored the effects of the surface-to-volume ratio on the sensitivity of a microsensor. We further examined the effects of each individual dimension on the sensitivity of a microsensor. We particularly ARTICLE IN PRESS Base r i, j Contacts Sensin g area 1 i2 n Unit block m 2 3 1 j Microwire Fig. 1. Schematic view of the relationship between a film sensor and individual unit blocks. a b t Fig. 2. The dimensions of a micropattern. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]]2 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 studied microwire sensors, whose lengths were much larger than the widths. A microsensor may also have a rectangular shape, i.e., the length is about the same as the width. If the length is large, the corresponding surface-to-volume ratio is larger than that of a microwire sensor. When the length is small, it is not easy to make a contact to the sensor. As such, microwire sensors became the focus of this work. 3. Fabrication of PPy and PEDOT–PSS microwires Conducting polymer micropatterns were generated using the ILL method [20–22] as follows (Fig. 3): (i) a layer of multiple conducting polymer coatings and a layer of a non-conducting polymer polymethyl methacrylate (PMMA) are heated up to the printing temperature, which is above the glass transition temperatures (T g ) of all polymers (Fig. 3a), (ii) a Si mold of desired patterns and the substrate are brought into physical contact by applied pressure, followed by subsequent cooling (Fig. 3b), and (iii) they are separated when their temperatures are below the lowest T g of all polymer materials, completing the pattern transfer from the mold to the conducting polymer layer (Fig. 3c). The three-step patterning process of the ILL is identical to that of the hot-embossing process [23]. The critical difference is that the substrate in the hot-embossing process has only the layer of the material to be printed, while the substrate in the ILL approach involves an additional intermediate layer of a non-conducting polymer. As a result of this difference, the conducting polymer patterns would be electrically isolated over the insulating intermediate layer, and patterns would be imprinted on the conducting polymer layer even if there were height differences existing between the features of the mold [20–22]. PMMA was chosen as the intermediate-layer material, because it is a good hot-embossing material. The PMMA has small thermal expansion coefficient of $5.0  10 À5 1C À1 and a small pressure shrinkage coefficient of $3.8  10 À7 psi À1 [24]. Its T g is around 105 1C. PPy (Sigma Aldrich Co.) and PEDOT–PSS (Baytron Co.) were used as received (5 wt% PPy in water and 1–1.4 wt% PEDOT–PPS in water) from the manufacturers. Their thin layers were generated by spin-coating the corresponding solutions on the PMMA sheet. Before coating the conducting polymers over the PMMA, all polymer solutions were kept in an ultrasonic bath for 1 h to remove any aggregate formation in solution from prolonged storage. The top surface of PMMA was treated with O 2 plasma (at 300 W watts for 45 s) to make it hydrophilic such that the water soluble conducting polymer solutions could be spin-coated over it. The key fabrication parameters in ILL are imprinting tempera- ture, imprinting force and imprinting time. The imprinting temperature was chosen to be higher than T g of PMMA and lower than T g of PPy and PEDOT–PSS in order to reduce thermal effects on these conducting polymers. The mold was slowly inserted into the substrate to avoid the dynamic effects in the embossed polymer. The silicon molds were fabricated using conventional ultraviolet lithography and deep reactive ion etch. The embossing temperature and pressure were 150 1C and 50 MPa, respectively. Fig. 4 shows a representative set of generated PPy microwires which have been used for sensing. Every sensor comprised six PPy or PEDOT–PSS microwires which were connected in parallel. Ag epoxy was placed at the two ends of these microwires as contact pads for electrical connection. The dimensions of micro- wires were changed to vary the surface-to-volume ratios of the microwires. One type of film and five types of microwire sensors were fabricated using the ILL method for either conducting polymer. Tables 1 and 2 give the corresponding dimensions of these sensors. ARTICLE IN PRESS Si mold Conducting polymer layer Intermediate polymer layer PMMA substrate Convex mold structure Concave mold structure Fig. 3. The three-step procedures to fabricate polymeric patterns using the proposed ILL method: (a) heating of the substrate, (b) insertion of the mold into the two polymer layers, and (c) separation of the mold and the substrate. Overall embossed area 100µm PMMA substrate PPy Fig. 4. (a) Perspective and (b) close-up (optical) views of PPy sensors generated on a PMMA sheet. Each PPy microwire has a width of 50 m m and a length of 2000 m m. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]] 3 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 4. Experimental setup for detection The experimental setup (Fig. 5) consisted of an air-tight chamber. All the tested sensors were placed at the same location inside the chamber, and the two contact wires for each sensor were taken out and connected to a Keithley probe station for I–V measurements. The humidity and temperature of the chamber were maintained at the room level and kept constant. After the chamber was closed, the sensor current was measured at 10 V to determine the base resistance. PPy and PEDOT–PSS microwires were exposed to methanol and acetone vapors, respectively, since they were sensitive to these two vapors, respectively. Methanol of a known volume was introduced into the chamber in a liquid form (as a droplet) using a micro-liter syringe. The same applied to acetone. The methanol droplet evaporated in 5–10 s. After the methanol droplets had evaporated completely, the current of a sensor was measured at 10 V continuously for 180s. Similarly, when acetone was introduced in the experimental chamber as a droplet, it evaporated in 2–4 s. After the methanol droplet had evaporated completely, the sensor current was monitored con- tinuously for 120 s. The observation time was reduced from 180 s for methanol to 120 s for acetone, since according to preliminary tests the PEDOT–PSS sensors responded to acetone exposure within 120 s. After a test, the chamber was purged by nitrogen and vented. For the next round of testing, the chamber was closed and the above procedure was repeated for detecting vapors of different concentrations. The masses of methanol and acetone were calculated from their known volumes (i.e., the evaporated volumes) and their densities at room temperature. The mass of air was calculated from the known volume (i.e., the volume of the chamber) and the density of air at room temperature. The concentration of the methanol was calculated from the ratio between the mass of methanol and that of air inside the test chamber. The same applied to acetone. The concentrations of methanol vapor ranged from 1.3 to 6.4 ppt. This range of methanol concentrations was about the same as the one reported in [8], which varied from about 1.5–5.0 ppt. The detection of methanol of lower concentra- tions (0.049–1.059 ppt) was reported in [13]. The acetone concentration of this work was varied from 0.6 to 5.8 ppt, which was lower than the concentration of 12.7 ppt considered in [7] (that is, 5% of acetone vapor pressure at 21 1C) and below the range of 104–416 ppt in [19]. 5. Sensing results 5.1. Exposure of PPy sensors to methanol vapor When the PPy film and microwire sensors were exposed to methanol vapor, response currents at 10 V varied with time in a wave-like form (Fig. 6). For the microwire sensors of different surface-to-volume ratios, the peak currents were reached be- tween 60 and 120 s after the methanol droplet had evaporated, and the response current varied between 1.5  10 À7 and 1.65 10 À7 A(Fig. 6a). Accordingly, the resistances varied between 6.67  10 7 and 5.99  10 7 O . Such a resistance included Ag–PPy–Ag contact resistance and intrinsic resistance of PPy wire. As indicated in [8], Ag–PPy–Ag contact was ohmic. The measured contact resistance was 1.21 Â10 5 O . Therefore, the contact resis- tance could be neglected, and the intrinsic resistance of the PPy wire dominated the detected resistance. The same applied to the case of PPy films. In addition, as examined in [25], the Ag–PEDOT/ PSS–Ag contact was also ohmic. The measured contact resistance ARTICLE IN PRESS Table 1 Dimensions of the PPy sensors used in the tests. PPy Width ( m m) Length ( m m) Thickness of PPy layer ( m m) Surface-to- volume ratio ( m m À1 ) Film 10,000 10,000 0.19 5.155 Microwire Type I 300 5000 0.19 5.162 Microwire Type II 100 5000 0.19 5.175 Microwire Type III 100 2000 0.19 5.176 Microwire Type IV 100 2000 0.13 7.655 Microwire Type V 100 2000 0.25 3.974 Table 2 Dimensions of the PEDOT–PSS sensors using in the tests. PEDOT–PSS Width ( m m) Length ( m m) Thickness of PEDOT–PSS layer ( m m) Surface-to- volume ratio ( m m À1 ) Film 10,000 10,000 0.30 3.334 Microwire Type I 300 5000 0.30 3.340 Microwire Type II 100 5000 0.30 3.353 Microwire Type III 100 2000 0.30 3.354 Microwire Type IV 100 2000 0.21 4.783 Microwire Type V 100 2000 1.15 0.890 N 2 inlet N 2 outlet Test chamber µL syringe Keithley probe station N 2 outlet N 2 inlet Microwire sensor Fig. 5. Experimental setup to determine the sensitivity of PPy and PEDOT–PSS sensors in detecting methanol and acetone, respectively. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]]4 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 was 5.00 Â10 3 O . Hence, the effect of contact resistance was also neglected in considering the PEDOT/PSS sensors. For the PPy film sensor, the response current reached a peak between 120 and 160 s, and the response current varied between 5 Â10 À7 and 5.29 Â10 À7 A(Fig. 6b). The corresponding resistances varied between 2.00 Â10 7 and 1.89  10 7 O . The transient nature of the response current was due to the fact that the methanol molecules were not stationary on the sensor causing the peak in the current. After the methanol droplet evaporated, it diffused inside the chamber and reached the sensors dynamically. The response current first increased and then decreased. The reason for this increase in response current may be attributed to the fact that methanol is a polar molecule which helps in interchain electron transfer in PPy. Also, the small size of the methanol molecules helped it to diffuse into the polymer chain more effectively, thus aiding conduction. The whole behavior was similar to the PPy response to humidity [18]. As the response current reached a maximum, the methanol vapor diffused out of PPy, since the methanol concentration in the PPy microwires was higher than that in the environment. This caused the decrease in the current. The time to reach the peak current was defined as the response time. Accordingly, microwire sensors have a shorter response time than film sensors. The same transient phenomenon of the response current was also found, for example, in [8] during the initial exposure of PPy films to methonal. The current had a rapid increase of more than two orders of magnitudes during the first 20 s of exposure and reached to a maximum value after 60 s [8]. The current settled down to a steady value below the maximum. Methanol was continuously supplied to a PPy sensor in [8]. This made the concentration of the methanol around the sensor was higher than that of our case, which did not provide continuous supply of the methanol. Therefore, the steady current obtained in [8] was much higher than the original current, while in our case the steady value was just a little higher than the original value. Therefore, the peak current was used in this work to calculate R exposure in determining the corresponding SI, since this gave a much larger SI compared with the case of adopting the steady current to calculate R exposure . Except for Type V microwires, the sensitivity increased in the order: FilmoType IoType IIoType IIIoType IV (Fig. 7). For the lowest methanol concentration of 1.3 ppt, the sensitivity of PPy film sensor was 1.6% for a surface-to-volume ratio of 5.155 m m À1 , as compared to PPy microwire Type IV with sensitivity of 36.44% for a surface-to-volume ratio of 7.655 m m À1 . Similarly, at the highest acetone concentration of 6.4 ppt, the PPy film sensitivity was 10.4% and Type IV microwire was 55.6%. These results indicate that in general the sensitivities of these sensors increase with the increasing surface-to-volume ratios. PPy film and microwires of Types I, II and III had the same thickness of 0.194 m m(Table 1). The sensitivities of the PPy film and microwire sensors had an approximately linear relationship with increasing methanol concentrations (Fig. 7). At the lowest methanol concentration of 1.3 ppt, the sensitivities of PPy film sensor were 1.6% and Type III microwire sensor was 8.2%. At the ARTICLE IN PRESS Time (s) 5.30 5.23 5.17 5.10 5.03 4.97 Time (s) 1.68 1.64 1.60 1.56 1.52 1.48 PPy microwire sensors PPy film sensor Response current (10 -7 A) Response current (10 -7 A) Fig. 6. Representative current responses of PPy (a) microwire and (b) film sensors during the 180-s exposure to methanol at a concentration of 3.8 ppt. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]] 5 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 highest methanol concentration of 6.4 ppt, the sensitivities of the film and Type III microwire sensors were 10.4% and 17.5%, respectively. The PPy thicknesses were varied for Types III, IV and V microwire sensors with their lengths and widths kept constant. This was done to study the effects of the PPy thicknesses on the sensitivity responses of the microwires. The thicknesses of the PPy layers were 0.131 and 0.253 m m for Types IV and V microwires, respectively. When the PPy thicknesses were varied, there were large variations in the surface-to-volume ratios. Type IV microwires had the highest surface-to-volume ratio and the highest sensitivities at all the methanol concentration levels (Fig. 7). The sensitivity of Type IV microwires at the lowest methanol concentration was 36.4% and at the highest concentra- tion was 55.6%. These results imply that for the PPy microwires their thicknesses may have larger effects on sensitivity than the length and width. It is also worth pointing out that, although the width of Type I wires was three times as large as that of Type II wires (they have the same length and width), their surface-to-volume ratios only differed by 0.013. Similarly, the 2.5-times difference in the widths between Types II and III led to only 0.001 difference in their surface-to-volume ratios. These two comparisons support the point raised in Section 2. That is, the changes of the length and width do not cause much change in the surface-to-volume ratio of a microsensor. However, it is interesting to see from Fig. 7 that these three types of microwires still had several percents of difference in their sensitivities of detecting methanol. It is noted that PPy sensors that other researchers used have demonstrated different sensitivities. For example, as indicated in [8], when exposed to 5 ppt of methanol, PPy films generated by inkjet printing [8] and dip-coating [12] have SI’s of 88% and 23%, respectively. For this concentration, the SI’s of our five types of sensors ranged from about 7–48%. The PPy films used to generate our sensors were spin-coated on substrates. Naturally, the sensitivity of a sensor should be affected by the sensing material used in the sensor. In addition to this, as indicated in [8], the way to make the film may also affect the sensitivity of a sensor. Manufacturing approaches affected the surface morphologies of generated films and subsequently the sensitivities of these films. For example, the inkjet-printed PPy film in [8] consisted of interconnected islands of average size 25 m m, while the spin- coated films have relatively flat surfaces. According to Eq. (1), the SI of the inkjet-printed PPy films may approximately equal that of the islands if each island is considered as a unit block of the film. Compared to a large film of flat surfaces, these small islands of the same thickness as the film have a higher surface-to-volume ratio. Therefore, their SI (and consequently the SI of the inkjet-printed film) should be higher than that of a spin-coated film. On the other hand, microstructures can be further generated in spin- coated films, functioning as sensing components and yielding higher sensitivities. This is implied by the different sensor responses of our five types of sensors. Thus, essentially, it should be feature sizes and shapes that affected the sensitivity of a sensor in addition to the sensing materials. 5.2. Exposure of PEDOT–PSS sensors to acetone vapor Fig. 8 shows the wave-like variation of the response current in detecting acetone using PEDOT–PSS sensors. The response current first decreased and then increased back to a steady value a little lower than its original value. At the initial stage, exposure of PEDOT–PSS to acetone reduced the conductivity of the PEDOT–PSS microwires. According to Ruangchuay et al. [26], acetone being a polar molecule, it dispersed inside the PPy matrix by hydrogen bonding. This mechanism disrupted the ordered structure and hence reduced the conductivity of PPy. A similar mechanism may be playing a role in reducing the conductivity of the PEDOT–PSS microwires in our case. Alternatively, acetone molecules diffused inside PEDOT–PSS, expanding the matrix, hindering the flow of charge carriers and thereby reducing conductivity of the micro- wires. As the response current reached a minimum, the acetone vapor diffused out of the PEDOT–PSS due to the fact that the acetone concentration in the PEDOT–PSS was higher than that in the environment. This caused the increase in the current. The sensing response of the PEDOT–PSS to acetone was different from that in the case when PPy sensors were used to detect methanol. The sensor current decreased to a minimum after about 90 s of exposure. The response times of film and microwire sensors were about the same. The same transient phenomenon was also found, for example, in [5] when the PEDOT–PSS sensors were employed to detect methanol and ethanol. However, due to the same reason addressed in Section 5.1, their steady currents were much different from the original currents, while in our case the steady value was just a little lower than the original value. Thus, the minimum current was used in this work to calculate R exposure in determining the corresponding SI, since this gave a much larger SI compared with the case of adopting the steady current to calculate R exposure . Except for Type V, the sensitivity of these sensors increased in the order: FilmoType IoType IIoType IIIoType IV. For the lowest acetone concentration of 0.64 ppt, the sensitivity of the film sensor was 0.05% for a surface-to-volume ratio of 3.33 m m À1 , as compared to Type IV microwires with sensitivity of 2.27% for a surface-to-volume ratio of 4.78 m m À1 . Similarly, at the highest acetone concentration of 5.8 ppt, the PEDOT–PSS film sensitivity was 0.5% and Type IV microwires was 20.6%. These results indicate that in general the sensitivities of these sensors increase with the increasing surface-to-volume ratios. On the other hand, as what we observed from the case of PPy detection, the large changes in widths and lengths among Types I, II and III microwires made only small changes in their surface-to-volume ratios. However, it can be seen from Fig. 9 that these three types of microwires also had several percents of difference in their sensitivities of detecting acetone. The thickness of the PEDOT–PSS layer for the film and Type I, II and III microwires was 0.3 m m(Table 2). The responses of the PEDOT–PSS microwires were more closely placed in the sensitiv- ity scale than the PPy microwires, while the overall trend was similar. For the PEDOT–PSS microwires (Fig. 9), the sensitivity increased from 0.05% for film sensor to 3% for Type III microwires, ARTICLE IN PRESS 60 55 50 45 40 35 30 25 20 15 10 5 0 1234567 Methanol concentration (ppth) Sensitivity (%) Type III Type II Type IV Type I; Type V Film Fig. 7. Sensitivity responses of the PPy sensors at various concentrations of methanol exposure. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]]6 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 at a concentration of 0.68 ppt and from 0.5% for film sensor to 20.7% for Type III microwires at a concentration of 5.8 ppt. The thickness of the PEDOT–PSS layer was varied with the length and width kept constant, similar to that in the PPy microwires. The thicknesses of the PEDOT–PSS layers were 0.21 m m and 1.15 m m for Types IV and V microwires, respectively. The sensitivity of Type IV microwires varied from 2.2% at 0.6 ppt to 20.6% at 5.8 ppt of acetone (Fig. 9). The sensitivities of Type III microwires were more than Type V and less than Type IV microwires. This trend is aligned with the increasing surface-to- volume ratios in order from Type V to III to IV. The sensitivities of these five types of sensors ranged from 0.05% to 20.7% when they were exposed to 0.6–5.8 ppt of acetone. Generally, they are higher than those (0.5–9.4%) reported in [7], which detected 12.7 ppt of acetone using 2.5-mm-wide composite films. The composite films consisted of PEDOT–PSS/insulating polymers or carbon black/insulating polymers. The difference in the sensitivities implies that both feature sizes and sensing materials affected the sensitivities. It is noted that gold–PEDOT/PSS–gold nanowires (8 m min length and 220 nm in diameter) were employed in [19] as sensors to detect acetone, whose concentrations ranged from 104 to 416 ppt. The corresponding sensitivities varied from 3% to 9%. The nanowires were synthesized using anodic aluminum oxide membranes. These results imply that our sensors also generally have higher sensitivities than the nanowire sensors. It has been indicated in [19] that, compared with film sensors, these nanowire sensors do not show higher sensitivities. They considered this was due to the impact of substrate roughness during film formation of the film sensors. In other words, different manufacturing approaches generate different features, making sensors have different sensitivities, as discussed in Section 5.1. In this work, ARTICLE IN PRESS Response current (10 -3 A) 0.60 0.58 0.56 0.52 0.48 0.50 0.66 0.64 0.62 Time (s) 0.54 PEDOT-PSS microwire sensors Response current (10 -3 A) 3.477 3.480 3.504 3.501 3.498 Time (s) 3.495 PEDOT-PSS film sensors 3.492 3.489 3.486 3.483 Fig. 8. Representative current responses of PEDOT–PSS (a) microwire and (b) film sensors during the 120-s exposure to acetone at a concentration of 5.8 ppt. 25 20 15 10 5 0 0.5 1.5 2.5 3.5 4.5 5.5 Acetone concentration (ppth) Sensitivity (%) Type IV Type III Type V Type II Type I Film Fig. 9. Sensitivity responses of the PEDOT–PSS sensors at various concentrations of acetone. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]] 7 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 the same manufacturing approach (as well as the same sensing material) has been used to generate the five types of sensors. Therefore, the manufacturing effect (as well as the sensing material) is not a concern here in comparing the sensitivities of these five types of sensors. 6. Statistical analysis of sensing data Based on the method of least square [27], a statistical program SAS has been run to fit the data points for further analyzing the sensing results and examining the effects of each individual dimension. We intended to find the relationship of the SI with the three geometrical dimensions and the vapor concentration. It was noticed that surface-to-volume ratio is a linear combination of the inverse of the three geometrical dimensions, and that the sensitivity should increase as this ratio increases. Therefore, we assumed that the SI was related with the inverse of these three dimensions. In addition, it was found from Figs. 7 and 9 that the SI had an approximately linear relationship with the vapor con- centration. Therefore, the SI was assumed to be directly related to the vapor concentration, instead of its higher orders. Let x 1 , x 2 , and x 3 represent the inverse of the length, the inverse of the width, and the inverse of the thickness of a microwire, respectively. Set x 4 to be the vapor concentration. y stood for the SI. Then, based on a linear aggregation model [28] and according to the dataset obtained in detecting methanol using PPy sensors, we got y ¼ 26479:7x 1 þ 268:7x 2 þ 9:9x 3 þ 1:9x 4 À 55:2. (2) The corresponding r 2 was 0.94. r 2 is the so-called coefficient of determination [26], and indicates how good the fitting is. It ranges from 0 to 1. The fitting is better as r 2 is closer to 1. In view of the dataset obtained in detecting acetone using PEDOT–PSS sensors, the following equation was found: y ¼ 17459:2x 1 þ 114:5x 2 þ 1:2x 3 þ 1:7x 4 À 10:8. (3) The related r 2 was 0.76. To see clearly the effects of each individual dimension on the sensitivity from the above two equations, let’s consider an example. In designing a conducting polymer microwire sensor, the initially chosen dimensions could be 1000 m m  100 m m  0.1 m m. Based on these dimensions, next we considered how the changes in these three dimensions affect the sensitivity. Let alternative length, width and thickness be 1000m,100n, and 0.1 l, respectively, where m, n and l were three positive constants and their values determine the final values of the three dimensions. Substituting the inverse of these three dimensions into Eqs. (2) and (3) for x 1 , x 2 , and x 3 , we had y ¼ 26:5=m þ 2:7=n þ 99=l þ 1:9x 4 À 55:2, y ¼ 17:5=m þ 1:1=n þ 12=l þ 1:7x 4 À 10:8. (4) Eq. (4) indicates that, for the detection of methanol using PPy microwire sensors, the change in thickness had more effects than the change in length on the sensitivity, while the latter had more effects than the change in width. For example, y increased by 13.25, 1.35, and 45.5, respectively, when we respectively set m, n and l to be 0.5. According to Eq. (4) 2 , the effects on the sensitivities of the PEDOT–PSS microwire sensors in detecting acetone were ordered from the highest to the lowest as: the change in length, the change in thickness, and the change in width. For example, y increased by 8.65, 0.55, and 6, respectively, when we respectively set m, n and l to be 0.5. As could be seen from these two relationships, the changes in the dimensions had more effects on the sensitivities of PPy microwire sensors than those on the sensitivities of PEDOT–PSS microwire sensors. Also, the degree of influence of each individual dimension might vary with different conducting polymer microwire sensors. As discussed in Section 2, the thickness affected surface-to- volume ratio much more than the length and the width. Also, in principle the sensitivity increased with the increasing surface-to- volume ratio. However, the two relationships given in Eq. (4) indicate that the length had the same order of effects as the thickness on the sensitivity. Therefore, in addition to the thickness, it is also important to reduce the length of a microwire for increasing the sensitivity. These two relationships also imply that the length had more effects than the width. To see this clearly, we compared the effects of the length with those of the width via the detection of acetone using PPy sensors. PPy film and Types I, II and IIII microwire sensors were chosen to detect acetone vapors, whose concentrations were 1.3, 3.2, 4.5, 5.8, and 7.1 ppt (Fig. 10). The same setup and testing procedure as described in Section 4 were used. The overall trend of the sensor responses was similar to what has been found in the previous two sets of experiments (Figs. 7 and 9). The sensitivity increased from 3.6% for film sensor to 13.0% for Type III microwires at a concentration of 1.3 ppt and from 10.1% for film sensor to 29.0% for Type III microwires at a concentration of 7.1 ppt. For a particular concentration, the sensitivity increased in the order: FilmoType IoType IIoType III. Since these sensors had the same thickness of 0.194 m m, we could directly compare the effects of the length and width. The corresponding fitting result was y ¼ 19301:5x 1 þ 920:0x 2 þ 2:1x 4 À 5:0. (5) The related r 2 is 0.94. Following the same line of reasoning that was used to obtain Eq. (4) from Eqs. (2) and (3), by Eq. (5) we had y ¼ 19:3=m þ 9:2=n þ 2:1x 4 À 5:0. (6) This equation indicates that, for the detection of acetone using PPy microwire sensors, the change in length had the same order of effects as the change in width. For example, y increased by 8.15 and 4.6, respectively when we, respectively, set m and n to be 0.5. The exact mechanism that caused the different effects of dimensions on the sensitivities is not clear. We speculate that it is related to the internal structures and orientations of PPy and PEDOT–PSS. For example, if the internal structure of a polymer is orientated upward, then its width should have more effect than the thickness, while the thickness of another polymer should be more important than the width in detection when its internal structure is pointed horizontally. We leave this to future investigation. 7. Summary and conclusions In this work, microwires of PPy and PEDOT–PSS were fabricated using the ILL technique. The microwires had different dimensions for achieving different surface-to-volume ratios. For PPy, the surface-to-volume ratio varied from 3.974 to 7.655 m m À1 . ARTICLE IN PRESS 0 5 10 15 20 25 30 35 1.3 Acetone cencentration ( pp th) Sensitivity (%) Type III Type II Type I Film 3.2 4.5 5.8 7.1 Fig. 10. Sensitivity responses of the PPy sensors at various concentrations of acetone. C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]]8 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 For PEDOT–PSS, the surface-to-volume ratio varied from 0.890 to 4.873 m m À1 . The PPy film and microwire sensors were exposed to methanol vapor whose concentrations ranged from 1.3 to 6.4 ppt. Methanol exposure increased the response current of the PPy sensors. The PEDOT–PSS film and microwire sensors were exposed to acetone vapor whose concentrations ranged from 0.6 to 5.8 ppt. The response current of the sensors was reduced upon exposure to acetone vapor. In general, the sensitivities of the sensors were found to increase with increasing surface-to-volume ratios at various concentrations of the methanol and acetone vapors. The sensitivity data obtained from experiments were analyzed with the aid of a statistical program, SAS. From the point view of surface-to-volume ratio, the thickness should affect the sensitivity much more than the other two dimensions. However, the three relationships obtained from three sets of experiments, respectively, indicate that the effects of the three geometrical dimensions on the sensitivity of a microwire sensor vary with the conducting polymer materials and the targets to detect. In other words, which dimension has more effects on sensitivity is case- dependent. Results presented in this work can be potentially used to aid in the design of conducting polymer microwire sensors of high sensitivity. Acknowledgements This work was supported in part through NSF–DMI-0508454, NSF/LEQSF(2006)-Pfund-53 and NSF-ECS-0529104 Grants. References [1] L. Rupprecht, Conductive Polymers and Plastics in Industrial Applications, Plastic Design Library, 1999. [2] B. Adhikari, S. Majumdar, Polymers in sensor applications, Progress in Polymer Science 29 (2004) 699–766. [3] S.T. McGovern, G.M. Spinks, G.G. Wallace, Micro-humidity sensors based on a processable polyaniline blend, Sensors and Actuators B 107 (2005) 657–665. [4] J.W. Gardner, P.N. Bartlett, A brief history of electronic noses, Sensors and Actuators B 18–19 (1994) 210–211. [5] M.F. Mabrook, C. Pearson, M.C. Petty, Inkjet-printed polymer films for the detection of organic vapors, IEEE Sensors Journal 6 (2006) 1435–1444. [6] S.V. Patel, M.W. Jenkins, R.C. Hughes, W.G. Yelton, A.J. Ricco, Differentiation of chemical components in a binary solvent vapor mixture using carbon/ polymer composite-based chemiresistor, Analytical Chemistry 72 (2000) 1532–1542. [7] G.A. Sotzing, S.M. Briglin, R.H. Grubbs, N.S. Lewis, Preparation and properties of vapor detector arrays formed from Poly(3,4-ethylenedioxy)thiophene- Poly(styrenesulphonate)/Insulating polymer composites, Analytical Chemis- try 72 (2000) 3181–3190. [8] M.F. Marbrook, C. Pearson, M.C. Petty, Inkjet printed PPy thin films for vapor sensing, Sensors and Actuators B 115 (2006) 547–551. [9] W.A. Daoud, J.H. Xin, Y.S. Szeto, Polyethylenedioxythiophene coatings for humidity, temperature and strain sensing polyamide fibers, Sensors and Actuators B 109 (2005) 329–333. [10] J.H. Cho, J.B. Yu, J.S. Kim, S.O. Sohn, D.D. Lee, J.S. Huh, Sensing behaviors of polypyrrole sensor under humidity condition, Sensors and Actuators B 108 (2005) 389–392. [11] C.P. de Melo, B.B. Neto, E.G. de Lima, L.F. B de Lira, J.E.C. de Souza, Use of conducting Polypyrrole blends as gas sensors, Sensors and Actuators B 109 (2005) 34 8–354. [12] H.K. Jun, Y.S. Huh, B.S. Lee, S.T. Lee, J.O. Lim, D.D. Lee, J.S. Huh, Electrical properties of polypyrrole gas sensors fabricated under various pre-treatment conditions, Sensors and Actuators B 96 (2003) 576–581. [13] L. Jiang, H.K. Jun, Y.S. Huh, J.O. Lim, D.D. Lee, J.S. Hoh, Sensing characteristics of polypyrrole-poly(vinyl alcohol) methanol sensors prepared by in situ vapor state polymerization, Sensors and Actuators B 105 (2005) 132–137. [14] Q. Ameer, S. Adeloju, Polypyrrole-based electronic noses for environmental and industrial analysis, Sensors and Actuators B 106 (2005) 541–552. [15] T. Yamauchi, K. Kojima, K. Oshima, M. Shimomura, S. Miyauchi, Glucose- sensing characteristics of conducting polymer bound with glucose oxidase, Synthetic Metals 102 (1999) 1320. [16] L. Ruangchuay, A. Sirivat, J. Schwank, Electrical conductivity response of polypyrrole to acetone vapor: effect of dopant anions and interaction mechanisms, Synthetic Metals 140 (2004) 15–21. [17] L.M. Torres-Rodriguez, M. Billon, A. Roget, G. Bidan, A Polypyrrole-biotin based biosensor: elaboration and characterization, Synthetic Metals 102 (1999) 1328–1329. [18] K. Suri, S. Annapoorni, A.K. Sarkar, R.P. Tandon, Gas and humidity sensors based on iron oxide-polypyrrole nanocomposites, Sensors and Actuators B 81 (2002) 277–282. [19] Y. Dan, Y. Cao, T.E. Mallouk, A.T. Johnson, S. Evoy, Dielectrophoretically assembled polymer nanowires for gas sensing, Sensors and Actuators B 125 (2007) 55–59. [20] C. Luo, R. Poddar, X. Liu, Innovative approach for replicating micropatterns in a conducting polymer, Journal of Vacuum Science and Technology B 24 (2006) L19–L22. [21] A. Chakraborty, X. Liu, C. Luo, An intermediate-layer lithography method for generating multiple microstructures made of different conducting polymers, Microsystem Technologies 13 (2007) 1175–1184. [22] X. Liu, A. Chakraborty, C. Luo, Generation of all-polymeric diodes and capacitors using an innovative intermediate-layer lithography, in: J.P. Martingale (Ed.), Progress in Solid State Electronics Research, Nova Science Publishers, Inc., 2007, pp. 127–139. [23] M. Hecklele, W. Bacher, K.D. Mu ¨ ller, Hot embossing-the molding technique for plastic microstructures, Microsystem Technologies 4 (1998) 122–124. [24] S.Y. Chou, P.R. Krauss, P.J. Renstorm, Nanoimprint Lithography, Journal of Vacuum Science and Technology B 14 (1996) 4129–4133. [25] Z. Liu, Y. Su, K. Varahramyan, Inkjet-printed silver conductors using silver nitrate ink and their electrical contacts with conducting polymers, Thin Solid Films 478 (2005) 275–279. [26] L. Ruangchuay, A. Sirivat, J. Schwank, Electrical conductivity response of polypyrrole to acetone vapor: effect of dopant anions and interaction mechanisms, Synthetic Metals 140 (2004) 15–21. [27] R.L. Plackett, The discovery of the method of least squares, Biometrika 59 (1972) 239–251. [28] E.F. Fama, J.D. MacBeth, Risk, return and equilibrium: empirical tests, Journal of Political Economy 71 (1973) 607–636. ARTICLE IN PRESS C. Luo, A. Chakraborty / Microelectronics Journal ] (]]]]) ]]]–]]] 9 Please cite this article as: C. Luo, A. Chakraborty, Effects of dimensions on the sensitivity of a conducting polymer microwire , Microelectron. J (2009), doi:10.1016/j.mejo.2008.11.064 . surface-to-volume ratios at various concentrations of the methanol and acetone vapors. The sensitivity data obtained from experiments were analyzed with the aid of a statistical. the ratio between the mass of methanol and that of air inside the test chamber. The same applied to acetone. The concentrations of methanol vapor ranged from